Sydney Morning Herald takes down opinion piece about AI after author used AI to write it

The Sydney Morning Herald has taken down an opinion piece defending the use of artificial intelligence at universities, after it emerged its author used AI to write it.

Cath Ellis is the pro vice chancellor of quality and integrity at Western Sydney University, and was featured on the newspaper's website arguing against concerns that generative AI was diminishing the value of higher education.

Sunday's column was a response to an earlier opinion piece by Macquarie University academic Kylie Moore-Gilbert.

In that article, published by the same masthead, Dr Moore-Gilbert accused Australia's universities of committing "widespread, industrial-scale fraud", and said workplaces would soon play host to a "real-time experiment in degree by [Chat]GPT".

In Professor Ellis's rebuttal, she wrote that she had recently spent time "at the centre of what generative AI is doing to higher education", and universities had been working to adapt to the technology.

"The [university] sector is not standing still. It is in the middle of a significant transformation," the now-retracted article read.

"What we are doing now … is shifting from assessing what students submit to assuring what they actually know, can do and have become."

A spokesperson for Western Sydney University (WSU) confirmed Professor Ellis had used generative AI, in what they said was a "sophisticated and appropriate" way.

"To write her opinion article, Prof. Ellis uploaded 40,000 words of her own original materials into a Copilot Large Language Model," a statement read.

"The model summarised her extensive base of knowledge, providing prompts.

"This was the basis of the early drafts, reflecting Prof. Ellis's own thinking, ideas and opinions built up over more than a decade of dedicated work as a global leader in this field."

WSU told the Herald that Professor Ellis shared her article with the university's media team, who also used AI tools to "suggest further improvements".

According to the editorial guidelines of Nine, publisher of the Herald, journalists and editorial employees are permitted to use AI tools where there is a "genuine benefit in doing so".

"This may include using assistive AI to do initial research, prompt ideas, analyse data, suggest headlines or key point summaries," the guidelines read.

"The use of assistive AI does not require declaration."

The guidelines maintain AI will not be used to write stories for publication, and that any task completed with the help of AI tools should be "critically scrutinised, verified and checked by a human" before it is used to create content.

'Clearly this is unacceptable'

Nine did not provide a statement to the ABC.

But a subsequent article published by the Herald on Wednesday quoted its editor, Jordan Baker, who said the piece had been retracted because it did not meet their editorial standards.

"The Herald was not informed of the use of AI in the compilation of the article by either the author or Western Sydney University," she said.

The Herald did not respond to the ABC's question on whether they would commission work by Professor Ellis in the future.

WSU said it was not aware that Professor Ellis's conduct was a breach of the Herald's editorial guidelines.

Professor of Journalism Practice at the University of Technology Sydney's Centre for Media Transition, Monica Attard, said Professor Ellis would have done better to declare her use of AI in the piece itself.

"Had the writer centred the use of AI to prove the point of what she was actually writing, that in her view there is some validity and efficacy in using generative AI, then there wouldn't have been a problem," she said.

Professor Attard said newsrooms were becoming increasingly alert to generated content.

"There have been … dozens of cases where people have used generative AI to actually literally generate an article, and then put that forward as their own work," she said.

But she said Australian editorial desks were still at an "experimental stage" with how generative AI fitted into their workflow.

"In other jurisdictions, for example in Scandinavia, it's much more widely used and they do have very, very sophisticated editorial processes in place to make sure that it is largely safe outward-facing usage," she said.

"But we're not there yet."